Evidence Explorer

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	15,059	30,035	126	210	8.37	6.99	1.38

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	165	209	300	711	880	2,177
Comparator	4	180	228	321	728	903	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	15,055	30,004	118	217	7.84	7.23	1.38

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	165	209	300	711	879	2,177
Comparator	2	180	227	320	727	903	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,072	10,072	24,541	24,706	227	203	9.25	8.22	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	171	387	784	1,617	1,774	2,188
Comparator	15	154	378	797	1,638	1,774	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,072	10,072	24,551	24,700	216	213	8.80	8.62	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	172	385	784	1,620	1,774	2,188
Comparator	4	153	378	798	1,637	1,771	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,065	10,065	11,861	11,260	83	105	7.00	9.32	1.50

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	14	174	209	301	714	888	2,177
Comparator	2	152	207	298	664	802	2,175

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,065	10,065	11,863	11,276	100	96	8.43	8.51	1.49

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	174	209	301	715	889	2,177
Comparator	4	152	207	298	665	803	2,175

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	31,036	66,663	292	503	9.41	7.55	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	165	378	773	1,620	1,774	2,189
Comparator	2	189	453	938	1,726	1,888	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	31,028	66,715	297	495	9.57	7.42	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	165	377	773	1,621	1,774	2,189
Comparator	4	189	455	941	1,726	1,887	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,069	10,069	11,885	11,351	86	101	7.24	8.90	1.51

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	168	209	301	719	888	2,177
Comparator	2	153	207	300	675	825	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,069	10,069	11,888	11,357	96	102	8.08	8.98	1.49

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	168	209	302	719	890	2,177
Comparator	4	152	207	300	676	825	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	15,059	30,035	126	210	8.37	6.99	1.38

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	165	209	300	711	880	2,177
Comparator	4	180	228	321	728	903	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	15,055	30,004	118	217	7.84	7.23	1.38

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	165	209	300	711	879	2,177
Comparator	2	180	227	320	727	903	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,063	10,063	24,615	24,414	215	222	8.73	9.09	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	12	171	389	789	1,622	1,777	2,189
Comparator	7	154	373	789	1,620	1,767	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
10,063	10,063	24,582	24,448	234	206	9.52	8.43	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	171	385	787	1,624	1,777	2,189
Comparator	4	154	375	794	1,617	1,766	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	31,028	66,715	297	495	9.57	7.42	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	165	377	773	1,621	1,774	2,189
Comparator	4	189	455	941	1,726	1,887	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
12,844	24,760	31,036	66,663	292	503	9.41	7.55	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	165	378	773	1,620	1,774	2,189
Comparator	2	189	453	938	1,726	1,888	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Adalimumab (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	15,442	25,932	128	166	8.29	6.40	1.40

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	209	300	710	877	2,177
Comparator	2	206	231	335	781	964	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	15,429	25,914	125	203	8.10	7.83	1.37

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	161	209	300	710	875	2,177
Comparator	5	203	231	334	781	963	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,411	9,411	23,463	22,995	201	199	8.57	8.65	1.32

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	179	399	804	1,652	1,799	2,189
Comparator	5	165	384	798	1,613	1,760	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,411	9,411	23,440	23,034	218	201	9.30	8.73	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	14	181	397	801	1,652	1,799	2,189
Comparator	9	167	389	801	1,614	1,762	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,412	9,412	11,305	10,917	85	87	7.52	7.97	1.53

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	14	182	209	305	734	903	2,177
Comparator	5	164	207	306	705	853	2,188

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,412	9,412	11,314	10,919	94	80	8.31	7.33	1.53

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	182	209	305	735	906	2,177
Comparator	11	165	207	307	706	852	2,188

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	31,567	59,259	303	450	9.60	7.59	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	375	760	1,610	1,763	2,189
Comparator	2	224	510	1,034	1,841	1,984	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	31,580	59,152	300	488	9.50	8.25	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	161	376	760	1,610	1,763	2,189
Comparator	5	222	507	1,034	1,840	1,985	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,418	9,418	11,223	10,891	95	79	8.46	7.25	1.53

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	177	209	302	724	901	2,177
Comparator	11	166	208	309	696	843	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,418	9,418	11,218	10,867	83	98	7.40	9.02	1.52

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	14	175	209	301	724	898	2,177
Comparator	5	165	208	307	692	843	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	15,429	25,914	125	203	8.10	7.83	1.37

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	161	209	300	710	875	2,177
Comparator	5	203	231	334	781	963	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	15,442	25,932	128	166	8.29	6.40	1.40

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	209	300	710	877	2,177
Comparator	2	206	231	335	781	964	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,408	9,408	23,460	22,896	216	202	9.21	8.82	1.32

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	14	178	399	804	1,655	1,803	2,189
Comparator	11	165	384	789	1,616	1,764	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
9,408	9,408	23,463	22,833	211	214	8.99	9.37	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	176	398	804	1,654	1,803	2,189
Comparator	5	165	383	784	1,615	1,762	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	31,567	59,259	303	450	9.60	7.59	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	375	760	1,610	1,763	2,189
Comparator	2	224	510	1,034	1,841	1,984	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RCT) vs. Etanercept (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
13,198	20,237	31,580	59,152	300	488	9.50	8.25	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	10	161	376	760	1,610	1,763	2,189
Comparator	5	222	507	1,034	1,840	1,985	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RCT)) and comparator (Etanercept (RCT)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	49,736	108,680	319	455	6.41	4.19	1.24

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	209	302	711	880	2,189
Comparator	2	201	232	339	802	990	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	49,706	108,643	316	502	6.36	4.62	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	209	302	711	878	2,189
Comparator	1	200	232	339	802	990	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,012	32,012	82,322	79,777	540	533	6.56	6.68	1.19

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	166	397	839	1,718	1,861	2,190
Comparator	2	141	374	802	1,683	1,835	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,012	32,012	82,356	79,854	552	520	6.70	6.51	1.19

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	166	397	839	1,719	1,862	2,190
Comparator	2	141	375	803	1,685	1,839	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,033	32,033	37,914	36,974	216	211	5.70	5.71	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	165	209	303	716	885	2,189
Comparator	2	140	207	305	697	848	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,033	32,033	37,908	36,945	212	223	5.59	6.04	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	209	303	717	885	2,189
Comparator	2	140	207	305	697	849	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	106,234	256,456	766	1,272	7.21	4.96	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	384	808	1,694	1,840	2,190
Comparator	2	221	539	1,127	1,942	2,062	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	106,154	256,282	757	1,267	7.13	4.94	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	384	808	1,693	1,839	2,190
Comparator	1	221	538	1,126	1,942	2,063	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,006	32,006	37,949	37,166	222	220	5.85	5.92	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	167	209	304	717	888	2,189
Comparator	2	143	207	307	698	857	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,006	32,006	37,943	37,137	202	219	5.32	5.90	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	166	209	304	717	888	2,189
Comparator	2	144	207	307	698	857	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	49,706	108,643	316	502	6.36	4.62	1.23

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	209	302	711	878	2,189
Comparator	1	200	232	339	802	990	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	49,736	108,680	319	455	6.41	4.19	1.24

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	209	302	711	880	2,189
Comparator	2	201	232	339	802	990	2,189

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,014	32,014	82,248	79,444	540	539	6.57	6.78	1.19

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	167	397	837	1,715	1,861	2,190
Comparator	2	140	368	798	1,673	1,813	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
32,014	32,014	82,293	79,553	544	540	6.61	6.79	1.19

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	167	397	836	1,719	1,863	2,190
Comparator	2	140	370	798	1,677	1,816	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	106,154	256,282	757	1,267	7.13	4.94	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	384	808	1,693	1,839	2,190
Comparator	1	221	538	1,126	1,942	2,063	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Etanercept (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
42,290	83,140	106,234	256,456	766	1,272	7.21	4.96	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	160	384	808	1,694	1,840	2,190
Comparator	2	221	539	1,127	1,942	2,062	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Etanercept (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	48,404	122,177	310	530	6.40	4.34	1.24

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	209	303	712	879	2,189
Comparator	1	187	231	325	726	906	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	48,387	122,130	297	555	6.14	4.54	1.24

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	164	209	303	712	878	2,189
Comparator	2	187	230	325	725	906	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,129	33,129	84,271	82,587	573	559	6.80	6.77	1.18

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	390	824	1,706	1,852	2,190
Comparator	1	146	369	806	1,687	1,841	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,129	33,129	84,280	82,556	546	575	6.48	6.96	1.18

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	164	391	826	1,706	1,849	2,190
Comparator	3	146	369	805	1,686	1,839	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,127	33,127	38,944	37,200	218	236	5.60	6.34	1.30

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	162	209	301	711	875	2,189
Comparator	3	145	207	304	662	810	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,127	33,127	38,943	37,228	245	216	6.29	5.80	1.30

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	209	302	711	875	2,189
Comparator	1	145	207	304	662	810	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	104,149	281,323	732	1,366	7.03	4.86	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	164	389	822	1,705	1,848	2,190
Comparator	2	198	474	998	1,796	1,959	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	104,221	281,443	751	1,336	7.21	4.75	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	389	822	1,705	1,852	2,190
Comparator	1	199	474	998	1,797	1,960	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,115	33,115	38,945	37,305	215	220	5.52	5.90	1.31

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	164	209	301	711	874	2,189
Comparator	2	144	208	303	665	819	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,115	33,115	38,932	37,319	242	219	6.22	5.87	1.30

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	209	301	711	873	2,189
Comparator	1	144	208	304	665	821	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	48,387	122,130	297	555	6.14	4.54	1.24

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	164	209	303	712	878	2,189
Comparator	2	187	230	325	725	906	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - Original Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	48,404	122,177	310	530	6.40	4.34	1.24

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	209	303	712	879	2,189
Comparator	1	187	231	325	726	906	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,121	33,121	84,219	82,673	549	564	6.52	6.82	1.18

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	164	389	825	1,706	1,852	2,190
Comparator	3	147	375	810	1,684	1,831	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
33,121	33,121	84,257	82,688	568	563	6.74	6.81	1.18

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	389	824	1,707	1,854	2,190
Comparator	1	146	375	810	1,685	1,831	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	104,149	281,323	732	1,366	7.03	4.86	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	2	164	389	822	1,705	1,848	2,190
Comparator	2	198	474	998	1,796	1,959	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Kaplan-Meier

Tofacitinib (RWD) vs. Adalimumab (RWD) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - New Exclusion Covariates, Open Claims]

Power

Target subjects	Comparator subjects	Target years	Comparator years	Target events	Comparator events	Target IR (per 1,000 PY)	Comparator IR (per 1,000 PY)	MDRR
41,071	99,812	104,221	281,443	751	1,336	7.21	4.75	1.14

Table 1a. Number of subjects, follow-up time (in years), number of outcome events, and event incidence rate (IR) per 1,000 patient years (PY) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after propensity score adjustment, as well as the minimum detectable relative risk (MDRR). Note that the IR does not account for any stratification.

Cohort	Min	P10	P25	Median	P75	P90	Max
Target	1	164	389	822	1,705	1,852	2,190
Comparator	1	199	474	998	1,797	1,960	2,190

Table 1b. Time (days) at risk distribution expressed as minimum (min), 25th percentile (P25), median, 75th percentile (P75), and maximum (max) in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) cohort after propensity score adjustment.

Attrition

Figure 1. Attrition diagram, showing the Number of subjectsin the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group after various stages in the analysis.

Population Characteristics

Table 2. Select characteristics before and after propensity score adjustment, showing the (weighted) percentage of subjects with the characteristics in the target (Tofacitinib (RWD)) and comparator (Adalimumab (RWD)) group, as well as the standardized difference of the means.

Propensity Scores

Figure 2. Preference score distribution. The preference score is a transformation of the propensity score that adjusts for differences in the sizes of the two treatment groups. A higher overlap indicates subjects in the two groups were more similar in terms of their predicted probability of receiving one treatment over the other.

Covariate Balance

Systematic Error

Figure 4. Systematic error. Effect size estimates for the negative controls (true hazard ratio = 1) and positive controls (true hazard ratio > 1), before and after calibration. Estimates below the diagonal dashed lines are statistically significant (alpha = 0.05) different from the true effect size. A well-calibrated estimator should have the true effect size within the 95 percent confidence interval 95 percent of times.

Evidence Explorer

Results Summary

Target vs. Comparator ~ Outcome [Analysis, DB]:

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [On Treatment (Stratification) - New Exclusion Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Matching) - Original Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - Original Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [Intent-to-Treat (Stratification) - Original Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Stroke (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores

Covariate Balance

Systematic Error

Kaplan-Meier

Tofacitinib (RCT) vs. Adalimumab (RCT) ~ Acute Myocardial Infarction (LEGEND Version) [On Treatment (Matching) - New Exclusion Covariates, Open Claims]

Power

Attrition

Population Characteristics

Propensity Scores